Null pointer panic

[aderouineau]

What version are you running?

8.21.1

Are you using Docker or Kubernetes to run your system?

Np

Are you running a single node or a cluster?

2-node cluster

What did you do?

1. Start node 1 with 10+ GB SQLite database
2. Start node 2, let the transfer and restore complete
3. Stop node 2 (without removing it from the cluster)
4. Start node 2 again
5. Run .nodes all before node 2 is restored

What did you expect to happen?

I expected node 2 to not panic.

What happened instead?

[store] 2024/02/23 15:19:33 opening store with node ID 2, listening on localhost:4012
[store] 2024/02/23 15:19:33 ensuring data directory exists at /home/derouin/rqlite/data2
[store] 2024/02/23 15:19:33 old snapshot directory /home/derouin/rqlite/data2/snapshots does not exist, nothing to upgrade
[snapshot-store] 2024/02/23 15:19:33 store initialized using /home/derouin/rqlite/data2/rsnapshots
[snapshot-store] 2024/02/23 15:19:33 checking consistency of snapshot store at /home/derouin/rqlite/data2/rsnapshots
[snapshot-store] 2024/02/23 15:19:33 check complete
[store] 2024/02/23 15:19:33 1 preexisting snapshots present
[store] 2024/02/23 15:19:33 first log index: 4, last log index: 4, last applied index: 4, last command log index: 0:
panic: runtime error: invalid memory address or nil pointer dereference
[signal SIGSEGV: segmentation violation code=0x1 addr=0x8 pc=0xf4a0e4]

goroutine 72 [running]:
github.com/hashicorp/raft.(*Raft).CommitIndex(...)
        /tmp/tmp.RHSvXLaim1/pkg/mod/github.com/hashicorp/raft@v1.6.1/api.go:1212
github.com/rqlite/rqlite/v8/store.(*Store).CommitIndex(0xc0000b0e60?)
        /tmp/tmp.RHSvXLaim1/src/github.com/rqlite/rqlite/store/store.go:859 +0x4
github.com/rqlite/rqlite/v8/cluster.(*Service).handleConn(0xc0001ba090, {0x18764d0?, 0xc00016cd90})
        /tmp/tmp.RHSvXLaim1/src/github.com/rqlite/rqlite/cluster/service.go:291 +0x207
created by github.com/rqlite/rqlite/v8/cluster.(*Service).serve in goroutine 14
        /tmp/tmp.RHSvXLaim1/src/github.com/rqlite/rqlite/cluster/service.go:228 +0x25

[otoolep]

When you run .nodes which node is rqlite connected to?

[aderouineau]

It's connected to node 1, but I'm not sure it makes a difference?

[otoolep]

Got it, thanks. I believe #1699 will fix this -- there should be more checks of the Store state in place.

[otoolep]

I'll aim to make a release with the fix in place within the next day or so.

[otoolep]

If you want to patch your own tree, and build locally, the linked PR will show you basically what you need to do.

[aderouineau]

No rush. So far I've just been messing around with rqlite and putting it through its paces, so-to-speak :)

[otoolep]

All good -- your work on rqlite has been very helpful.